MDM RNA-seq Analysis Report
RNAseq data from Adam Fields’ Laboratory, University of California San Diego (UCSD).
1 Summary
Number of samples: 180Groups: HIV- moderate cannabis, HIV- daily cannabis, HIV- naive to cannabis, HIV+ naive to cannabis, HIV+ daily cannabis, HIV+ moderate cannabisTreatments: CBD, CBDpIL1B, IL1B, THC, THCpIL1B, VehicleHIV status: HIVn, HIVpCannabis use: moderate, daily, naive
2 Workflow Overview
We processed the RNA-seq data from monocyte-derived macrophages (MDMs) using the following pipeline:
treatment: vehicle, IL1b, CBD, THC, CBD+IL1B, THC+IL1b
hiv_status: HIV- vs HIV+
cannabis: naive, moderate, daily
Briefly. differential expression (DE) analyses were performed using DESeq2 with a factorial design including HIV status, cannabis use, and their interaction. This model allowed estimation of HIV-associated transcriptional changes separately in cannabis-naive and cannabis-using participants, and formal testing of effect modification by cannabis use. For comparison, stratified analyses restricted to HIV-positive participants were also conducted to estimate cannabis-associated effects within this subgroup.
| Model | Uses HIV–? | Uses HIV+? | Tests HIV effect | Tests cannabis effect | Tests interaction | Power |
|---|---|---|---|---|---|---|
~ cannabis * hiv_status |
✅ | ✅ | ✅ | ✅ | ✅ | ⭐⭐⭐ |
~ cannabis + hiv_status |
✅ | ✅ | ✅ | ✅ | ❌ | ⭐⭐⭐⭐ (if no interaction) |
HIV+ only: ~ cannabis |
❌ | ✅ | ❌ | ✅ (within HIV+) | ❌ | ⭐⭐ |
3 Summary Analyses
3.1 DESeq2 Models used below
The differential expression model used:
\[ design(dds) ~ treatment + HIV status + cannabis \]
This is an additive model without interaction. This initial model assumes:
- HIV effect is the same regardless of cannabis
- Cannabis effect is the same regardless of HIV
- No interaction allowed
This allows us to investigate the acute transcriptional effect of treatment, accounting for HIV and cannabis background for example.
3.2 Volcano Plots
Warning: Using `size` aesthetic for lines was deprecated in ggplot2 3.4.0.
ℹ Please use `linewidth` instead.
ℹ The deprecated feature was likely used in the EnhancedVolcano package.
Please report the issue to the authors.
Warning: The `size` argument of `element_line()` is deprecated as of ggplot2 3.4.0.
ℹ Please use the `linewidth` argument instead.
ℹ The deprecated feature was likely used in the EnhancedVolcano package.
Please report the issue to the authors.
volcano_list[["cannabis_daily_vs_naive"]]3.3 PCA Plots
using ntop=500 top features by variance
using ntop=500 top features by variance
using ntop=500 top features by variance
using ntop=500 top features by variance
3.4 Top Genes Tables
# DT::datatable(top20_list[["cannabis_daily_vs_naive"]], options = list(pageLength = 5, scrollX = TRUE), rownames = FALSE)
DT::datatable(top20_list[["cannabis_daily_vs_naive"]]|>
dplyr::mutate(across(where(base::is.numeric), ~ round(.x, 20))),
rownames = FALSE,
extensions = 'Buttons',
options = list(
dom = 'Bfrtip',
buttons = c('copy', 'csv', 'excel'),
pageLength = 10
)) - Only genes with padj < 0.05 and |log2FC| > 1 are labeled in volcano plots.
- PCA plots can be faceted by
hiv_statusto observe group separation.
- Interactive tables allow scrolling and sorting top genes for each contrast.
3.5 Pathway enrichment analysis
Gene Ontology (GO) enrichment analysis was performed separately for upregulated and downregulated genes for each contrast. Genes with an adjusted p-value < 0.01 and absolute log₂ fold change > 2 were selected. Enrichment analysis was conducted using the clusterProfiler package with the Biological Process (BP) ontology, using all expressed genes as the background universe. Enrichment results were visualized using dot plots, and leading-edge genes contributing to each enriched pathway were extracted for downstream interpretation.
Dot plots showing Gene Ontology Biological Process enrichment for upregulated (left) and downregulated (right) genes across experimental contrasts. Genes were considered differentially expressed if they exhibited an adjusted p-value < 0.01 and an absolute log₂ fold change > 2. Dot size represents the number of genes contributing to each pathway, and color indicates the adjusted p-value. Enrichment analyses were performed using the full set of expressed genes as background. Pathways are shown separately for genes increased and decreased in expression to facilitate biological interpretation.
NULL
NULL
NULL
NULL
NULL
NULL
NULL
NULL
3.6 Top Differentially Expressed Genes
- Heatmap showing variance-stabilized expression (VST) of the top N differentially expressed genes for the indicated contrast. Genes were selected based on adjusted p-value and absolute log2 fold change from DESeq2 analysis. Rows represent genes (labeled by gene symbol), and columns represent samples. Expression values are row-scaled (z-score) to emphasize relative expression patterns. Samples are ordered by HIV status, cannabis exposure, and treatment condition.
3.7 Selected / Candidate Genes
- Heatmap showing variance-stabilized expression (VST) of selected genes of interest across samples. Genes were chosen a priori based on biological relevance (e.g., HIV response, inflammation, cannabinoid signaling). Rows represent genes (gene symbols), and columns represent samples. Expression values are row-scaled to highlight relative differences across conditions. Samples are ordered by HIV status, cannabis exposure, and treatment.
4 Impact of HIV on inflammatory phenotype
4.1 Methods
Here we compare PWH vs PWoH within vehicle-treated samples and adjust fo cannabis use.
4.2 Differentially Expressed Genes
Volcano plot of DE genes between PWH and PWoH macrophages. TNF is the only significantly upregulated inflammatory gene (padj < 0.05).
4.3 Enrichment of inflammatory pathways in vehicle-treated macrophages from PWH
Over-representation analysis (ORA) of genes differentially expressed between PWH and PWoH macrophages under vehicle conditions highlights enrichment of pathways related to cytokine production, adaptive immune responses, and immune effector processes. Point size indicates the number of DE genes in each pathway, while color represents the significance of enrichment (-log10 adjusted p-value)
4.4 GSEA of inflammatory pathways in vehicle-treated macrophages.
Ridgeplot from GSEA showing the distribution of pathway genes across the ranked gene list; peaks to the right indicate upregulation in PWH
Dot size indicates pathway size (number of genes). NES = normalized enrichment score; positive values indicate upregulation in PWH. Red dots indicate pathways where TNF is part of the core enrichment contributing to the NES. All three pathways are significantly enriched (NES 2.17–2.23, p.adjust < 0.01).
5 Modulation of IL1-B response by Cannabis Use
IL1B induces a strong inflammatory response in PWH naive to cannabis, with upregulation of classic cytokine and immune effector genes (e.g., IL1B, IL6, TNFAIP6). Moderate and daily cannabis use appears to dampen this response, as most genes are not significantly induced in these groups. This suggests cannabis exposure may modulate macrophage inflammatory activation.
5.1 Method
5.1.1 Model rationale
The interaction framework allows us to distinguish:
- Main effect of cannabis: baseline differences in gene expression between cannabis exposure groups
- Main effect of treatment: transcriptional changes induced by IL1B stimulation
- Interaction effect: whether the effect of IL1B stimulation differs depending on cannabis exposure. The interaction term specifically tests whether the IL1B-induced transcriptional response is modified by cannabis use, rather than assuming identical treatment effects across exposure groups.
5.1.2 Statistical model
For each gene (i), counts were modeled using a negative binomial distribution:
\[ log(μi)=β0+β1⋅Cannabis+β2⋅Treatment+β3⋅(Cannabis×Treatment) \]
Where:
- μi is the expected expression of gene (i)
- β0 is the intercept
- β1 represents the main effect of cannabis exposure
- β2 represents the main effect of IL1B treatment
- β3 represents the interaction effect, capturing differential IL1B responses by cannabis status
5.1.3 Approach
RNA-seq data from MDM of PWH are analyzed using DESeq2. Genes responsive to IL1B stimulation are filtered to retain only those significant in at least one cannabis use condition (adjusted p < 0.05). Log2 fold-change (log2FC) values are extracted for naive, moderate, and daily cannabid exposure. Heatmaps were generated using ComplexHeatmap, with non-significant log2FC values desaturated and significant changes indicated with an asterisk (*). Key inflammatory genes (e.g., IL1B, TNF, CXCL10) were highlighted in bold. Interactive tables were produced with DT, combining log2FC values and significance for each condition.
5.2 Results
Heatmap showing log2 fold-change (log2FC) of IL1B-responsive inflammatory genes across three cannabis conditions: naive , moderate, and daily. Only genes significant in at least one condition (adjusted p < 0.05) are included. Log2FC values for non-significant conditions are shown but visually desaturated, while significant changes are highlighted with an asterisk (*). Blue, white, and red indicate negative, no change, and positive log2FC, respectively. Gray squares represent missing values (NA). Rows are clustered based on expression patterns. The log2FC color legend is shown on the right, and significance is indicated in a separate legend.
6 IL1B-induced inflammatory response modulated by CBD or THC
Here we compared inflammatory response triggered by IL-1B alone, IL-1B + THC and IL-B+CBD and adjusted for HIV status in persons Cannabis Naive.
6.1 1:1 Comparison IL-1B ± THC or CBD versus Vehicle
6.2 1:1 Comparison IL-1B modulation by THC or CBD
7 Cannabis * HIV Status Interaction models.
7.1 In HIV+ vs HIV- individuals
7.1.1 DESeq2 Model
The differential expression model used:
\[ design(dds\ in\ vehicle) <- ~ cannabis * HIV\ status \]
Using ALL samples, with:
- a main effect of HIV status
- a main effect of cannabis
- an interaction term (does cannabis modify the HIV effect?)
7.1.2 Results
Under vehicle conditions, TNF expression in macrophages from PWH was higher than in PWoH at the canonical transcript level (log₂FC ≈ +2.9), consistent with modest basal inflammatory activation. However, this difference did not reach statistical significance after multiple testing correction (FDR ≈ 1.0).
7.2 In HIV+ individuals, does moderate cannabis reduce TNF compared to naive?
7.2.1 DESeq2 Model
The differential expression model used:
\[ design(dds\ in\ Vehicle\ for\ HIV+) <- ~ cannabis \]
7.2.2 HIV+ moderate vs naive
7.2.3 HIV+ daily vs naive
7.2.4 HIV+ daily+moderate vs naive
In HIV-positive macrophages under vehicle conditions, cannabis use was associated with a dose-dependent reduction in TNF expression, with moderate and daily users showing markedly lower TNF levels compared to cannabis-naive individuals (log₂FC −3.6 and −4.1, respectively; nominal p < 0.05).
7.3 TNF summary
7.3.1 Canonical TNF Transcripts
7.3.2 TNF expression in macrophages by HIV status and cannabis use.
Normalized RNA-seq counts for all TNF transcripts (all ENSEMBL IDs annotated as TNF) were summed per sample and plotted as mean ± 95% confidence interval. PWH = people with HIV; PWoH = people without HIV. Bars show TNF expression under vehicle conditions stratified by cannabis use (naive, moderate, daily).
8 Gene set enrichment analysis (GSEA).
8.1 GSEA for Inflammatory pathway activation in HIV+ macrophages.
In vehicle-treated MDMs, are inflammatory pathways upregulated in HIV+ compared to HIV−, adjusting for cannabis use?
8.1.1 Method
- Genes were ranked by log₂ fold change from DESeq2 differential expression analysis comparing HIV-positive to HIV-negative macrophages under vehicle conditions. GSEA was performed using the clusterProfiler package with Hallmark inflammatory gene sets from MSigDB. Statistical significance was assessed using permutation-based testing, and results are reported as normalized enrichment scores (NES) with Benjamini–Hochberg adjusted P values.
8.1.2 DESeq2 Model
The differential expression model used:
\[ design(dds\ in\ vehicle) <- ~ cannabis * HIV\ status \] Using ALL samples with:
- a main effect of HIV status
- a main effect of cannabis
- an interaction term (does cannabis modify the HIV effect?)
8.1.3 Results
Volcano plot showing the genes belonging to the selected pathway.
Warning: ggrepel: 23 unlabeled data points (too many overlaps). Consider
increasing max.overlaps
Warning: ggrepel: 28 unlabeled data points (too many overlaps). Consider
increasing max.overlaps
8.1.4 Interpretation
Gene set enrichment analysis revealed significant enrichment of TNFα signaling via NF-κB (NES = 1.92, adjusted P = 6.7×10⁻⁶) and a broader inflammatory response signature (NES = 1.47, adjusted P = 0.009) in HIV-positive macrophages relative to HIV-negative controls. Core enrichment genes included canonical inflammatory mediators such as TNF, IL1B, IL6, CXCL10, and NFKBIA, indicating coordinated activation of cytokine-driven inflammatory programs. NES stands for Normalized Enrichment Score..
Vehicle-treated macrophages from PWH exhibited strong enrichment of TNFα/NFκB and inflammatory response pathways relative to HIV− controls, a signature that was attenuated in cannabis-exposed cells.
8.2 GSEA for cannabis effect within HIV+
Does cannabis dampen inflammatory signaling specifically in HIV+ MDMs? In other words, how cannabis moderate use changes gene expression specifically in HIV+ macrophages, relative to HIV+ naïve cannabis users.
8.2.1 Results
8.2.1.1 GSEA
8.2.1.2 Volcano of selected genes
Volcano plot showing the genes belonging to the selected pathway.
- 👉 TNFα / inflammatory genes are systematically downregulated in HIV+ MDMs with moderate cannabis use compared to HIV+ naïve users.
- 👉 Cannabis use dampens the basal inflammatory transcriptional program in HIV+ macrophages, particularly TNFα/NF-κB–driven signaling.
8.3 GSEA for interaction alone: Mechanistic ++
Are inflammatory pathways differentially regulated by cannabis in HIV+ vs HIV−? Is cannabis modulation different in HIV+ vs HIV−?
- 👉 The suppressive effect of cannabis on inflammatory genes is significantly stronger in HIV+ macrophages than in HIV− macrophages.
- 👉 Cannabis dampens inflammatory signaling preferentially in HIV+ cells.
9 Are HIV-induced transcriptomic changes attenuated (or altered) in cannabis users?
- We explore the following contrasts:
| Result | Meaning |
|---|---|
res_hiv_naive |
HIV effect in cannabis-naive |
res_hiv_users |
HIV effect in cannabis users |
res_interaction |
Difference between the two |
The differential expression model used:
\[ design(dds\ in \ vehicle) <- ~ cannabis\ (binary) + HIV\ status + cannabis\ (binary):HIV\ status \]
9.1 HIV effect among cannabis-naive individuals
9.2 HIV effect among cannabis-users
9.3 Difference in HIV effect between cannabis users and naive
9.4 GSEA: Does cannabis attenuate HIV response?
9.4.1 Approach
Scatter plots show the log2 fold change of HIV-associated gene expression in cannabis-naive MDMs (x-axis) versus cannabis-using MDMs (y-axis). Each facet represents a selected Gene Ontology Biological Process (GO term) enriched for genes showing interaction effects between HIV and cannabis.
Points belowthe dashed identity line (y = x) indicate genes whose HIV effect is reduced in cannabis users compared with naive individuals (attenuation).Points abovethe line indicate genes with enhanced HIV effect in users.Red pointshighlight genes belonging to the specific GO term of the facet.Gene labelsindicate the top genes per GO term showing the largest reduction in HIV-induced expression (largest difference lfc_naive – lfc_users).
The differential expression model used:
\[ design(dds\ in \ vehicle) <- ~ cannabis\ (binary) + HIV\ status + cannabis\ (binary):HIV\ status \] Equivalent to
\[ design(dds\ in \ vehicle) <- ~ cannabis\ (binary)*HIV\ status \] Using ALL samples with:
- a main effect of HIV status
- a main effect of cannabis
- an interaction term (does cannabis modify the HIV effect?)
| Term | Biological meaning |
|---|---|
(Intercept) |
Baseline (HIV–, non-cannabis) |
hiv_status_HIVp_vs_HIVn |
HIV effect in non-cannabis users |
cannabis_bin_moderate_daily |
Cannabis effect in HIV– |
cannabis_binmoderate_daily:hiv_statusHIVp |
Modification of HIV effect by cannabis |
Hence combined contrasts list(c("hiv_status_HIVp_vs_HIVn","cannabis_binmoderate_daily.hiv_statusHIVp")) is HIV effect in cannabis users.
9.4.2 Results for selected inflammatory genes
| Quadrant | x = lfc_naive | y = lfc_users | Interpretation |
|---|---|---|---|
Top-right |
> 0 | > 0 | HIV upregulated in both groups |
Bottom-right |
> 0 | < 0 | HIV up in naive, down or attenuated in users |
Bottom-left |
< 0 | < 0 | HIV downregulated in both groups |
Top-left |
< 0 | > 0 | HIV down in naive, up in users (interaction effect) |
9.4.3 Results by GO terms
- Genes falling below the identity line exhibit reduced HIV-associated induction in cannabis users compared with cannabis-naive individuals, consistent with attenuation of HIV-driven macrophage activation.
- This visualization demonstrates that cannabis partially attenuates HIV-driven activation of immune and translational pathways, with most pathway genes falling below the identity line, highlighting a coherent dampening effect on macrophage activation.
- Although cannabis use did not induce a global reversal of HIV-associated transcriptional changes, interaction-based GSEA revealed significant attenuation of innate immune and translational programs, suggesting a partial dampening of HIV-driven macrophage activation.
10 Inflammasome Pathways
10.1 GO terms and canonical inflammasome complex
10.2 Executive summary
This analysis investigated the transcriptional landscape of 30 vehicle-treated Monocyte-Derived Macrophages (MDMs). The central finding is that HIV status drives a robust pro-inflammatory program, primarily through the TNF\(\alpha\)/NF\(\kappa\)B axis. However, daily cannabis use significantly attenuates (dampens) this induction. While individual gene variance in primary human cells is high, pathway-level analysis (GSEA) and interaction modeling provide definitive evidence that cannabis modulates the immune response by “blunting” the HIV-induced inflammatory spike.
10.3 Methodology & Quality Control
Differential expression was modeled using DESeq2 with the design ~ cannabis * hiv_status. This allows for the extraction of:
The HIV Main Effect: The impact of HIV in cannabis-naive individuals.
The Interaction Effect: How the HIV response changes in moderate vs. daily cannabis users.
10.4 Diagnostic Evaluation of Dispersion and Outliers
10.4.1 Dispersion
Genes followed the expected mean-variance trend. Highly inflammatory genes showed high dispersion, typical for the “bursty” nature of cytokine transcription in primary MDMs.
- Red line: The “trend” or “expected” dispersion based on the gene’s mean expression.
- Blue circles: The final values used by DESeq2 for the Wald test (the black dots “shrunken” toward the red line).
- Blue circles with rings: “Outliers” that have such high variance they were not shrunken toward the line.
- In a healthy experiment, the black dots should cluster tightly around the red line.
10.4.2 Outlier Detection (Cook’s Distance)
All samples for the key lead IL1B showed Cook’s distances \(< 0.1\), confirming that the observed trends are not driven by rogue samples.
10.4.3 Coefficient of Variation (CV)
Inflammasome genes showed high biological heterogeneity (CV for AIM2 = 1.64; IL1B = 1.17). This justifies the use of GSEA, which aggregates signal across pathways to overcome single-gene noise.
10.5 Results: The HIV-Induced Inflammatory Signal
In cannabis-naive individuals, HIV infection triggers a strong “Priming” signal.
10.5.1 The HIV Effect Volcano Plot (Naive Group)
This Volcano shows IL1B as a high-magnitude outlier on the X-axis. Even with \(padj > 0.05\), the biological magnitude (\(LFC > 2\)) is substantial.
10.5.2 Inflammmasome Heatmap
Visualizes the raw “heat” of these 9 genes across all samples, showing that the HIVn vs HIVp contrast is strongest in the Naive donor subset.
10.6 Results: Cannabis-Mediated Dampening (Interaction)
| Component | Log2FC | Biological Interpretation |
|---|---|---|
| HIV Effect (Naive) | 2.18 | HIV ‘turns on’ IL1B priming significantly. |
| Interaction (Daily) | -1.75 | Cannabis ‘pulls down’ the HIV response in daily users. |
| Net HIV Effect (Daily) | 0.43 | The remaining HIV signal is reduced to near-baseline levels. |
The core of the study is the Interaction Term, which measures if cannabis users “react” differently to HIV. This “Net Effect” calculation shows that while HIV is still present, the daily cannabis usage has neutralized roughly 80% of the transcriptional priming for IL1B.
10.6.1 Interaction Volcano Plot
10.6.2 Heatmap
10.6.3 GSEA Enrichment Plot: Pathway-Level Suppression
To validate the dampening effect of cannabis on HIV-induced inflammation, we employed both pathway-level and gene-specific visualizations.
To move beyond single-gene analysis, we performed Gene Set Enrichment Analysis (GSEA) on the interaction model. We identified a significant suppression of the HALLMARK_TNFA_SIGNALING_VIA_NFKB gene set (p = 0.000566). The characteristic downward enrichment curve demonstrates that daily cannabis use systematically attenuates the core transcriptional machinery responsible for HIV-induced inflammation. This pathway-level evidence confirms that the dampening observed in \(IL1B\) is part of a broader regulatory shift, effectively ‘cooling’ the hyper-inflammatory state typically induced by chronic HIV infection
GSEA analysis confirmed a significant global suppression of the TNF\(\alpha\)/NF\(\kappa\)B signaling pathway, a master regulator of the macrophage inflammatory response.
10.6.4 LFC-LFC Comparison Plot: Global vs. Specific Modulation
This pathway-level shift is further detailed in the LFC-LFC Comparison Plot (Plot 2), which reveals that key inflammasome mediators—specifically IL1B and NLRP3—deviate significantly from the identity line. These genes occupy the ‘dampened’ quadrant, demonstrating that while HIV provides a robust priming signal in naive users, this response is consistently attenuated in daily cannabis users.
10.6.5 Summary table
10.7 Results: Focus on IL1B: The Overlap Node
10.7.1 IL1B Grouped Boxplot:
Shows the “Step-up” in counts from HIV- to HIV+ is large in Naive, smaller in Moderate, and nearly flat in Daily.
10.7.2 Comparison Scatter Plot:
The LFC-LFC comparison plot demonstrates a global trend of inflammatory dampening in daily cannabis users. Specifically, the key inflammasome mediator IL1B deviates significantly from the identity line (diagonal), shifting from a robust induction in naive users (\(LFC = 2.18\)) to a near-baseline state in daily users (\(LFC = 0.43\)).
10.8 Inflammasome Core: Donor Heterogeneity
10.8.1 IL1-B only
To visualize the donor-to-donor variability that explains your high adjusted p-value (\(padj = 0.93\)), we can look at the expression of IL1B across every single one of your 30 vehicle samples.
10.8.2 More genes
10.9 Interpretation (AI assisted)
The inflammasome functions as a two-stage biological alarm. IL-1B acts as the “Ammunition” of this system; it is strictly regulated and typically kept “unloaded” to prevent collateral tissue damage. Chronic HIV infection serves as a potent Priming Signal (Signal 1), forcing the cell to produce a massive surplus of \(IL1B\) transcripts, essentially “loading the gun” for an inflammatory explosion. In contrast, genes like CASP1 and PYCARD represent the “Machinery”—the firing pin and barrel—which the cell keeps at a steady state of readiness. Our data shows that while HIV significantly increases the “Ammunition” (\(IL1B\) Log2FC = +2.18), daily cannabis use effectively “unloads” the system, dampening this priming signal and reducing the net inflammatory potential back toward baseline levels.
| Feature | IL-1B (The Ammunition) | CASP1 / PYCARD (The Machinery) |
|---|---|---|
| Analogy | The Gunpowder/Payload | The Firing Pin/Trigger |
| Baseline State | Strictly 'OFF' (Silenced) | Always 'ON' (Constitutive) |
| HIV Response (Naive) | High Induction (Log2FC +2.18) | Stable/Low Induction (Log2FC < 1.0) |
| Cannabis Interaction | Targeted Dampening (Log2FC -1.75) | Minimal change (Stable) |
| Clinical Significance | Controls magnitude of inflammation | Controls readiness to respond |